Rob Leclerc: Safety Orgs' High p(doom) Means They'll Sandbag and Veto Every Model Release
robleclerc · x · 2026-09-15
Quoting news that OpenAI wants UKAISI to play a central role in the UK's regulatory framework, Rob Leclerc argues that even nominally independent safety bodies aren't unbiased: their employees generally carry a high p(doom) and will see sandbagging and vetoing every release as their duty, pointing to Anthropic having once deemed GPT-2 too dangerous to release. 'Nothing will ever be safe enough,' he writes.
More from AGI Musings
- Gary Marcus: We must think about AI in terms of collaborating to make the world better — GaryMarcus · 2026-09-15
- Synthesia CEO: the more human AI feels, the more real human time will be worth — alexvoica · 2026-09-15
- Columbia economist Dave Holtz takes leave to lead AGI economics research at OpenAI — daveholtz · 2026-09-15
- Chris Manning: the AI race debate boils down to 'better us than them' — chrmanning · 2026-09-15
- New essay asks: why humans have consciousness but AI cannot — AnnaCiaunica · 2026-09-15
- AI risk discourse is split into two camps, and the smart take is the secret third thing — arpitingle · 2026-09-15